philbert440/Qwen3.6-27B-Uncensored-Aggressive
philbert440/Qwen3.6-27B-Uncensored-Aggressive is a BF16 full-precision checkpoint derived from the Qwen3.6-27B vision-language model. This model features heavy "abliteration" to achieve a high degree of openness (0.417), making it permissive for cyber and NSFW prompts while maintaining refusal for illegal or harmful content. It is designed for users requiring an uncensored Qwen3.6 variant for further quantization or direct use on capable GPUs.
Loading preview...
Model Overview
philbert440/Qwen3.6-27B-Uncensored-Aggressive is an "abliterated" version of the Qwen/Qwen3.6-27B vision-language model. This specific variant is presented as a BF16 full-precision source checkpoint, intended for users who plan to perform further quantization or have GPUs with sufficient headroom to run it directly. It is part of a larger collection of uncensored Qwen3.6-27B models.
Key Characteristics
- Aggressive Abliteration: This model represents the most open tier within its family, with an "openness" score of 0.417. This means it is highly permissive for prompts related to cyber activities, hacking, malware, lock-picking, and NSFW content.
- BF16 Precision: Provided as a full-precision checkpoint, offering the highest fidelity for subsequent processing or direct deployment.
- Selective Uncensoring: While "uncensored" in the context of cyber and NSFW content, the model retains safety refusals for prompts concerning weapons, drugs, political persuasion, surveillance, and extremism.
Use Cases
This model is particularly suited for:
- Research and Development: Exploring the behavior of highly open language models in specific domains.
- Content Generation: Applications requiring less restrictive content generation, particularly in cyber-related or NSFW contexts, where ethical and lawful use is maintained by the developer.
- Custom Fine-tuning: Serving as a robust base for further fine-tuning or quantization to meet specific application requirements.